How to download large file with binary mode in python?

Question

I am code a download function in python. The file size >1GB. The server is linux, HTTP server is Karrigell. Client is browse, Firefox or IE. I meet a big trouble.

At first, I use sys.stdout() to send file content.

file = open(path, 'rb')
size = os.path.getsize(path)

RESPONSE['Pragma'] = 'public'
RESPONSE['Expires'] = '0'
RESPONSE['Cache-Control'] = 'must-revalidate, pre-check=0'
RESPONSE['Content-Disposition'] = 'attachment; filename="' + os.path.basename(path) + '"'
RESPONSE['Content-type'] = "application/octet-stream"
RESPONSE['Content-Transfer-Encoding'] = 'binary'
RESPONSE['Content-length'] = str(os.path.getsize(path))

sys.stdout.flush()
chunk_size = 10000
handle = open(path, "rb")
while True:
    buffer = handle.read(chunk_size)
    if buffer:
        STDOUT(buffer)
    else:
        break
sys.stdout.flush()

The problem is the server out of memory! I know, stdout write content to memory first, then memory send to socket.

So, I modify the function. Send content to socket directly. I use the py-sendfile module. http://code.google.com/p/py-sendfile/

file = open(path, 'rb')
size = os.path.getsize(path)

sock = REQUEST_HANDLER.sock
sock.sendall("""HTTP/1.1 200 OK\r\nPragma: no-cache\r\nExpires: 0\r\nCache-Control: no-cache, no-store\r\nContent-Disposition: attachment; filename="%s"\r\nContent-Type: application/octet-stream\r\nContent-Length: %u\r\nContent-Range: bytes 0-4096/%u\r\nLocation: "%s"\r\n\r\n""" % (os.path.basename(path), size, size, os.path.basename(path)))

offset = 0
nbytes = 4096
while 1:
    try:
        sent = sendfile.sendfile(sock.fileno(), file.fileno(), offset, nbytes)
    except OSError, err:
        if err.errno in (errno.EAGAIN, errno.EBUSY):  # retry
            continue
        raise
    else:
        if sent == 0:
            break    # done
        offset += sent

This time, the server memory is OK, but browse die! The browse memory rise quickly! Not free until the socket accept whole file content.

I don't know how to deal with these problems. I think the second idea is right, send content to socket directly. But why browse can't free memory while accept data?

Answer 1

You should try to download the file in chunks. This is an example that works for me using urllib2

import os
import urllib2
import math

def downloadChunks(url):
    """Helper to download large files
        the only arg is a url
       this file will go to a temp directory
       the file will also be downloaded
       in chunks and print out how much remains
    """

    baseFile = os.path.basename(url)

    #move the file to a more uniq path
    os.umask(0002)
    temp_path = "/tmp/"
    try:
        file = os.path.join(temp_path,baseFile)

        req = urllib2.urlopen(url)
        total_size = int(req.info().getheader('Content-Length').strip())
        downloaded = 0
        CHUNK = 256 * 10240
        with open(file, 'wb') as fp:
            while True:
                chunk = req.read(CHUNK)
                downloaded += len(chunk)
                print math.floor( (downloaded / total_size) * 100 )
                if not chunk: break
                fp.write(chunk)
    except urllib2.HTTPError, e:
        print "HTTP Error:",e.code , url
        return False
    except urllib2.URLError, e:
        print "URL Error:",e.reason , url
        return False

    return file